Mojo function 🡭

render_layers

fn def render_layers(plots: List[AnyChart]) -> Canvas

Render every Plot in plots onto one shared coordinate system: one combined x/y domain across every layer, one set of axes/gridlines/ ticks, each mark drawn over the last in the order given.

Takes the marks that place their data on a continuous x/y axis in the caller’s own units, which is what one shared domain can mean: Mark.POINT, LINE, AREA, EFFECT_SCATTER, KDE, RUG, BARBS, TRICONTOUR, TRICONTOURF, TRIPLOT and TRIPCOLOR – so a rug under a density curve, two KDEs compared on one frame, isolines over a filled contour, or a scatter over a triangulated field all draw. Mark.BAR layers with identical ordered categories share a categorical frame and occupy adjacent subbands. One Mark.GROUPED_BAR or Mark.STACKED_BAR layer can instead own the whole category band. POINT/LINE/AREA layers align to category centers. These dispatch to _render_bar_combo_layers.

Two or more Mark.ARC plots draw concentric rings, plots[0] outermost. Each ring keeps its own proportions and palette; legend entries are prefixed with its one-based ring number. Ring geometry is assigned by the composition, so every input must use pie()’s default inner_radius_fraction=0. A single ARC layer renders exactly as that standalone pie or donut. All export backends use this same layout.

Every other mark raises with the reason: a categorical, polar, hierarchical or edge-list mark has no continuous x to share; Mark.CONTOUR/CONTOURF lay out in unpadded grid-index units rather than the data’s own coordinates; Mark.SINGLE_AXIS pins its points at the plot rect’s vertical midpoint, which encodes nothing. render_facets() has no allow-list at all and takes any mark, so it is what to reach for when a composition here is refused.

A Plot.secondary_axis() layer scales against its own y-domain on the right edge – not a Mark.RUG layer, which has no y dimension to scale. A Mark.POINT/EFFECT_SCATTER layer may use color/ color_categories/size encoding with its own scales and legend section; sections stack in one column in layer order. There is no per-series legend for flat-colored layers, and the field marks draw no color bar here, as they draw none standalone.

Domains combine across every layer: the x-axis spans all of them, and a layer whose y is measured from a baseline – Mark.AREA’s fill height, Mark.KDE’s density – forces zero into its axis group’s domain, the rule one Mark.AREA layer has always applied. Plot.scale_x_log()/scale_y_log() apply only to Mark.POINT/ LINE/AREA/EFFECT_SCATTER, the same restriction a standalone render has. A stack of nothing but Mark.RUG layers has no y-axis at all and draws none, as a standalone rugplot() does.

Shared chrome (background, gridlines, axis colors, margins, font size, Plot.labels() titles) comes from plots[0]; every other layer’s Theme governs only its own mark (mark_color, point_radius, line_width, line_smoothing, scaled by its own Theme.scale). A secondary-axis layer’s y_title captions the right axis.

Every Plot must share the same .size(); an empty list raises. Supersampled by plots[0].settings.theme.raster_supersample like render(), bumping every layer’s scale together on a copy (plots is a plain borrow).

Args:

• plots (List[AnyChart])

Returns:

Canvas

Raises:

fn def render_layers[*Cs: ChartLike](*charts: *Cs.values) -> Canvas

render_layers(plots) for charts named one by one, of any mark types: render_layers(bars, line). Each is erased for the shared list; the layout is the same.

Parameters:

• *Cs (ChartLike)

Args:

• *charts (*Cs.values)

Returns:

Canvas

Raises: